From a40a7b03fbc8404888549e71800f7f44bbf2ed1f Mon Sep 17 00:00:00 2001 From: Daniel Han Date: Mon, 1 Dec 2025 00:01:07 -0800 Subject: [PATCH] Typos --- unsloth/models/falcon_h1.py | 2 +- unsloth/models/gemma.py | 2 +- unsloth/models/gemma2.py | 2 +- unsloth/models/llama.py | 2 +- unsloth/models/mistral.py | 2 +- unsloth/models/qwen2.py | 2 +- unsloth/models/qwen3.py | 2 +- unsloth/models/qwen3_moe.py | 2 +- 8 files changed, 8 insertions(+), 8 deletions(-) diff --git a/unsloth/models/falcon_h1.py b/unsloth/models/falcon_h1.py index c6a413ee40..105ad6b9a0 100644 --- a/unsloth/models/falcon_h1.py +++ b/unsloth/models/falcon_h1.py @@ -726,7 +726,7 @@ class FastFalconH1Model(FastLlamaModel): # Solves https://github.com/unslothai/unsloth/issues/168 # Static KV Cache was introduced in 4.38.0, causing training to be much slower. - # Inferene can now be CUDAGraphed, but we shall retain the old rotary embeddings. + # Inference can now be CUDAGraphed, but we shall retain the old rotary embeddings. # https://github.com/huggingface/transformers/pull/27931 # https://github.com/huggingface/transformers/blob/v4.37.2/src/transformers/models/llama/modeling_llama.py import transformers.models.falcon_h1.modeling_falcon_h1 diff --git a/unsloth/models/gemma.py b/unsloth/models/gemma.py index cf5bdd8ebe..f22b358db8 100644 --- a/unsloth/models/gemma.py +++ b/unsloth/models/gemma.py @@ -416,7 +416,7 @@ class FastGemmaModel(FastLlamaModel): # Solves https://github.com/unslothai/unsloth/issues/168 # Static KV Cache was introduced in 4.38.0, causing training to be much slower. - # Inferene can now be CUDAGraphed, but we shall retain the old rotary embeddings. + # Inference can now be CUDAGraphed, but we shall retain the old rotary embeddings. # https://github.com/huggingface/transformers/pull/27931 # https://github.com/huggingface/transformers/blob/v4.37.2/src/transformers/models/llama/modeling_llama.py import transformers.models.gemma.modeling_gemma diff --git a/unsloth/models/gemma2.py b/unsloth/models/gemma2.py index de70f78736..234fc21000 100644 --- a/unsloth/models/gemma2.py +++ b/unsloth/models/gemma2.py @@ -550,7 +550,7 @@ class FastGemma2Model(FastLlamaModel): # Solves https://github.com/unslothai/unsloth/issues/168 # Static KV Cache was introduced in 4.38.0, causing training to be much slower. - # Inferene can now be CUDAGraphed, but we shall retain the old rotary embeddings. + # Inference can now be CUDAGraphed, but we shall retain the old rotary embeddings. # https://github.com/huggingface/transformers/pull/27931 # https://github.com/huggingface/transformers/blob/v4.37.2/src/transformers/models/llama/modeling_llama.py import transformers.models.gemma2.modeling_gemma2 diff --git a/unsloth/models/llama.py b/unsloth/models/llama.py index f6a0fff117..1fe4fbffbe 100644 --- a/unsloth/models/llama.py +++ b/unsloth/models/llama.py @@ -1533,7 +1533,7 @@ def PeftModel_fast_forward( # Solves https://github.com/unslothai/unsloth/issues/168 # Static KV Cache was introduced in 4.38.0, causing training to be much slower. -# Inferene can now be CUDAGraphed, but we shall retain the old rotary embeddings. +# Inference can now be CUDAGraphed, but we shall retain the old rotary embeddings. # https://github.com/huggingface/transformers/pull/27931 # https://github.com/huggingface/transformers/blob/v4.37.2/src/transformers/models/llama/modeling_llama.py class LlamaRotaryEmbedding(torch.nn.Module): diff --git a/unsloth/models/mistral.py b/unsloth/models/mistral.py index 1945e6a745..3ccd234a06 100644 --- a/unsloth/models/mistral.py +++ b/unsloth/models/mistral.py @@ -439,7 +439,7 @@ class FastMistralModel(FastLlamaModel): # Solves https://github.com/unslothai/unsloth/issues/168 # Static KV Cache was introduced in 4.38.0, causing training to be much slower. - # Inferene can now be CUDAGraphed, but we shall retain the old rotary embeddings. + # Inference can now be CUDAGraphed, but we shall retain the old rotary embeddings. # https://github.com/huggingface/transformers/pull/27931 # https://github.com/huggingface/transformers/blob/v4.37.2/src/transformers/models/llama/modeling_llama.py import transformers.models.mistral.modeling_mistral diff --git a/unsloth/models/qwen2.py b/unsloth/models/qwen2.py index 64ed707fe5..3f819d6dc1 100644 --- a/unsloth/models/qwen2.py +++ b/unsloth/models/qwen2.py @@ -60,7 +60,7 @@ class FastQwen2Model(FastLlamaModel): # Solves https://github.com/unslothai/unsloth/issues/168 # Static KV Cache was introduced in 4.38.0, causing training to be much slower. - # Inferene can now be CUDAGraphed, but we shall retain the old rotary embeddings. + # Inference can now be CUDAGraphed, but we shall retain the old rotary embeddings. # https://github.com/huggingface/transformers/pull/27931 # https://github.com/huggingface/transformers/blob/v4.37.2/src/transformers/models/llama/modeling_llama.py import transformers.models.qwen2.modeling_qwen2 diff --git a/unsloth/models/qwen3.py b/unsloth/models/qwen3.py index 1a0d641552..6f41579d88 100644 --- a/unsloth/models/qwen3.py +++ b/unsloth/models/qwen3.py @@ -451,7 +451,7 @@ class FastQwen3Model(FastLlamaModel): # Solves https://github.com/unslothai/unsloth/issues/168 # Static KV Cache was introduced in 4.38.0, causing training to be much slower. - # Inferene can now be CUDAGraphed, but we shall retain the old rotary embeddings. + # Inference can now be CUDAGraphed, but we shall retain the old rotary embeddings. # https://github.com/huggingface/transformers/pull/27931 # https://github.com/huggingface/transformers/blob/v4.37.2/src/transformers/models/llama/modeling_llama.py import transformers.models.qwen3.modeling_qwen3 diff --git a/unsloth/models/qwen3_moe.py b/unsloth/models/qwen3_moe.py index f0b2b811ba..94633ca2d4 100644 --- a/unsloth/models/qwen3_moe.py +++ b/unsloth/models/qwen3_moe.py @@ -200,7 +200,7 @@ class FastQwen3MoeModel(FastQwen3Model): # Solves https://github.com/unslothai/unsloth/issues/168 # Static KV Cache was introduced in 4.38.0, causing training to be much slower. - # Inferene can now be CUDAGraphed, but we shall retain the old rotary embeddings. + # Inference can now be CUDAGraphed, but we shall retain the old rotary embeddings. # https://github.com/huggingface/transformers/pull/27931 # https://github.com/huggingface/transformers/blob/v4.37.2/src/transformers/models/llama/modeling_llama.py\ import transformers.models.qwen3_moe.modeling_qwen3_moe